Towards shell model interactions with credible uncertainties
Oliver C. Gorton, Konstantinos Kravvaris · 2025
Read Towards shell model interactions with credible uncertainties by Oliver C. Gorton, Konstantinos Kravvaris (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-11889.txt.
Introduction
Oliver Gorton and Konstantinos Kravvaris build a way to fit the terms of the nuclear shell model to data, with trust-worthy bounds on the doubt. Their paper is from two thousand twenty-five.
The shell model foretells how nuclei are built, but its terms are fit to lab data. To reach far nuclei, by the drip lines, fits must be fast and their doubt true. Past ways cost too much and can play the doubt down. The team joins a fast stand-in, built by a stretch method, with a chance-based sampler. The stand-in remakes a known shell case with scant error. The sampler then fits the terms and gauges their doubt in full.
Catalog record
| CREATOR | Oliver C. Gorton, Konstantinos Kravvaris |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-11889
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Towards shell model interactions with credible uncertainties Oliver C. Gorton, Konstantinos Kravvaris Abstract Background The nuclear shell model is a powerful framework for predicting nuclear structure observables, but relies on interaction matrix elements fit to experimental data as its inputs. Extending the shell model’s applicability, particularly toward dripline nuclei, requires efficient fitting methods and credible uncertainty quantification. Traditional approaches face computational challenges and may underestimate uncertainties. Purpose We develop and test a framework combining eigenvector continuation and Markov Chain Monte Carlo to efficiently fit shell model interaction matrix elements and quantify their uncertainties. Methods Eigenvector continuation is used to emulate shell model calculations, reducing computational costs. The emulator enables Markov chain Monte Carlo sampling to optimize interaction matrix elements and rigorously assess parametric uncertainties. The framework is benchmarked using the USDB interaction in the sd -shell. Results The emulator reproduces the USDB interaction with negligible error, validating its use in shell model fitting applications. However, we find that to obtain credible predictive intervals, the model defect of the shell model itself, rather than experimental or emulator error, must be taken into account in order to obtain credible uncertainties. Conclusions The proposed framework provides an efficient and rigorous approach for fitting shell model interactions and quantifying uncertainties. Further, the normality assumption used in the past appears sufficient to describe the distribution of interaction matrix elements. However, it is crucial to account for model correlations to avoid underestimating uncertainties. I Introduction For the past 75 years, the shell model (SM) [ 1 , 2 , 3 , 4 ] has served as a foundational tool for accurately predicting nuclear structure observables. These include binding energies, spectra, electromagnetic and weak decays, as well as level density information, which can all be used to inform applications in nuclear astrophysics and nuclear technologies. The SM’s success has led to the development of various approaches aimed at extending its applicability, not only to unexplored regions of the nuclear chart but also to the study of collective phenomena [ 5 , 6 ] , a prospect that would naively go against the single-particle understanding that the SM offers. At its core, the SM describes the nucleus using a handful of degrees of freedom, namely valence protons and neutrons in a mean field, interacting with two-particle forces. How exactly the many-body problem that includes nucleon-nucleon and many-nucleon interactions can be reduced to such a description is a matter of active research [ 7 , 8 , 9 , 10 , 11 , 12 ] . Despite these first-principles efforts, the most accurate SM predictions typically rely on inputs derived from direct fits to experimental data [ 13 , 14 , 15 , 16 , 17 , 18 , 19 ] . The computational cost of these fits has inhibited rigorous error analysis. Recent progress has enabled rudimentary uncertainty quantification (UQ) approaches [ 20 , 21 , 22 ] , though these are currently restricted to relatively small-scale calculations. Consequently, there is a pressing need to develop a framework that can efficiently fit phenomenological SM interactions to experimental data while simultaneously providing credible uncertainties. Fitting shell model interactions directly translates to fitting the mean-field energies and residual two-body matrix elements. One first defines the valence (or active) space of single particle levels that the valence nucleons can occupy. The effective coupling constants of the interaction, collectively labeled by \bm{x} , include the single-particle energies (SPEs) and two-body matrix elements (TBMEs) of and between the SM orbitals. The total interaction with one- and two-body terms can be written as [ 15 ] : \hat{H}(\bm{x})=\sum_{i}\epsilon_{i}\hat{n}_{i}+\sum_{i\leq j,k\leq l;JT}V_{ijkl;JT}\hat{T}_{ijkl;JT} (1) where the indices i,j,k,l label the single-particle orbits (typically harmonic oscillator states); the collective index i is short for all quantum numbers defining an orbit (n_{i},l_{i},j_{i}) , with principle quantum numbers n_{i} , orbital angular momentum l_{i} , and total angular momentum j_{i} . The number operator for a given shell i is represented by \hat{n}_{i} and \hat{T} is the scalar two-body density operator [ 15 ] . The interaction matrix elements \bm{x} are shorthand for the full set of SPEs and TBMEs: x:(\epsilon_{i},V_{ijkl;JT}) , which must be obtained from either theoretical or experimental constrains. For a given parametrization of the interaction, \bm{x} , the eigenfunctions \ket{\Psi(\bm x)} and eigenenergies E(\bm{x}) are found by solving the time-independent Schrodinger equation: \hat{H}(\bm{x})\ket{\Psi(\bm x)}=E(\bm{x})\ket{\Psi(\bm x)}. (2) In the full configuration interaction (FCI) framework with a frozen core, equation ( 1 ) is typically restricted to a finite basis of Slater determinants. This approach represents the Hamiltonian as a matrix, whose eigenpairs correspond to the nuclear wave functions and energies. SM interactions, such as the USD-family (USD [ 23 , 13 ] , USDA/B [ 15 ] , USDC [ 19 ] ) have been highly successful in reproducing experimental binding energies, low-lying excitation energies, and transition probabilities. For sd -shell nuclei, the most common variants, USDA and USDB, consist of three single-particle energies and 63 two-body matrix elements. These matrix elements were empirically fit with specific linear combinations of the 66 parameters iteratively updated to minimize a \chi^{2} fit to 608 energy levels across 77 nuclei [ 15 ] . This procedure is remarkably efficient, requiring around 30 iterations, with each iteration taking 12 hours. It achieved a local optimum while keeping computational demands manageable, but provided limited insight into the robustness of the predictions. The goals of this paper are: 1) to benchmark a new, computationally efficient methodology for fitting SM interactions, which is critical for generating uncertainty-quantified interactions beyond the sd -shell–particularly in neutron-rich model spaces that are essential for studying r-process nucleosynthesis; and 2) to rigorously assess the parametric uncertainty and covariances of the phenomenological SM. While previous studies have investigated the uncertainty of the USDB interaction [ 20 , 21 ] and the SM more broadly [ 22 ] , these efforts have been limited to approximate statistical methods due to computational costs. Recent developments in computational methods offer promising solutions to these goals. Eigenvector continuation (EC) has emerged as a powerful tool for emulating solutions to parametric Hamiltonians. By constructing a reduced basis from known eigenstates, EC circumvents the need for repeated diagonalization of large Hamiltonian matrices, dramatically reducing computational costs. This acceleration makes advanced fitting algorithms, such as Markov Chain Monte Carlo (MCMC), feasible for fitting SM interactions. MCMC also provides a rigorous framework for exploring parameter uncertainties in nonlinear models, overcoming the limitations of traditional \chi^{2} minimization techniques. In this paper, we develop and test a framework combining EC and MCMC to efficiently fit SM interactions and obtain credible uncertainties. Using the sd -shell as a benchmark, we demonstrate that EC can reproduce the well-known USDB interaction with negligible error, validating its use as an emulator for SM calculations. (See Section II and Section III.1 .) We then employ MCMC to generate uncertainty-quantified interactions, investigating concerns of nonlinearities and correlations in the parameter space that were previously neglected. (See Section III.2 .) Finally, we propose a practical approach to correct these underestimations, ensuring that predictive intervals better reflect the true uncertainty of SM calculations. II Methods In this section, we review some aspects of the FCI SM and describe the algorithms employed to fit the interaction matrix elements. Next,